Artificial intelligence applications in the agriculture sector

The agricultural sector today, needs innovative approaches that increase productivity due to the increasing world population, decreasing agricultural land and the effects of climate change. Smart agriculture offers a modern solution that optimizes these agricultural processes through the integration of advanced technologies. Sensors, artificial intelligence, image processing algorithms and robotic systems play critical roles in increasing agricultural productivity as well as ensuring food safety. In particular, wheat classification has a strategic importance both in terms of preserving genetic diversity and developing agricultural practices suitable for environmental conditions. In this study, the performance of various deep learning models was analyzed in detail in order to ensure accurate and fast classification of wheat varieties. Models used; VGG19, InceptionV3, ResNet50, Xception, ResNet101, EfficientNetB5 and EfficientNetB7 and training, test accuracy rates are 96.44%, 97.48%, 95.77%, 94.92%, 95.59%, 97.61% and 95.42%, respectively. The findings revealed that EfficientNetB5 (97.61%) and InceptionV3 (97.48%) models were superior to the other models in terms of wheat grading performance. This success demonstrates that the EfficientNetB5 architecture provides better performance with optimized parameter structure and low computational costs. In addition, the superiority of the InceptionV3 model in identifying complex wheat patterns proves that the model can be used effectively in the field of agricultural image processing. In addition, it was determined that a Ensemble Learning model (98.2%) created by combining these two most successful models gave more successful results than all other models. In particular, the size and variety of agricultural images have been a decisive factor on the performance of the models. The results reveal that accurate wheat classification can minimize economic losses by optimizing the use of resources in agricultural production processes and contribute to the trade of agricultural products in accordance with international standards. This study aims to increase the effectiveness of smart agricultural practices by providing artificial intelligence-based solutions to agricultural classification processes.